Direct Measurement of the Visible to UV Photodissociation Processes for the PhotoCORM TryptoCORM

Abstract PhotoCORMs are light‐triggered compounds that release CO for medical applications. Here, we apply laser spectroscopy in the gas phase to TryptoCORM, a known photoCORM that has been shown to destroy Escherichia coli upon visible‐light activation. Our experiments allow us to map TryptoCORM's photochemistry across a wide wavelength range by using novel laser‐interfaced mass spectrometry (LIMS). LIMS provides the intrinsic absorption spectrum of the photoCORM along with the production spectra of all of its ionic photoproducts for the first time. Importantly, the photoproduct spectra directly reveal the optimum wavelengths for maximizing CO ejection, and the extent to which CO ejection is compromised at redder wavelengths. A series of comparative studies were performed on TryptoCORM‐CH3CN which exists in dynamic equilibrium with TryptoCORM in solution. Our measurements allow us to conclude that the presence of the labile CH3CN facilitates CO release over a wider wavelength range. This work demonstrates the potential of LIMS as a new methodology for assessing active agent release (e.g. CO, NO, H2S) from light‐activated prodrugs.


Introduction
While the toxicityo fc arbon monoxidei sw ell known, its biological significance and therapeutic properties have been increasingly recognized over recenty ears. [1,2] At low concentrations, CO is ap owerful anti-inflammatory and organ transplantation agentt hat also has significant potential as ac hemotherapy and anti-microbialp harmaceutical. [3][4][5][6][7][8] Controlled CO delivery remains as ignificant challenge, however,f or its practical medicinal use. One solution to delivering localized CO involves the use of carbon monoxide releasing molecules (CORMs) that exhibit CO release only when triggered. [1,2,[9][10][11][12] Transition metal carbonyl complexes that eject CO upon light activation (photo-CORMs) have emerged as the most promising delivery agents to date. [13][14][15] Most photoCORMsd eveloped so far incorporate group 7a nd 8t ransition metalsa nd require UV light to trigger CO release. [16] These wavelengths presents as ignificant problem for wide-scale photoCORM application as they penetrate tissue poorly,a nd am ajor challenge in this fielda tt he present time is therefore developmento fn ew visible-light activated photoCORMs. [17] The rational design of future photoCORMs could be considerably improved through obtaining am ore complete understandingo ftheir fundamentalp hotomolecular properties. [18] A number of groups are using this developmental approach, combining known transition metal photochemistry with density functional theory calculations to predict and interpret trends for trial compounds. [19][20][21] However, progress is currently hampered by as parsity of experimental measurements on photo-CORMst hat can be straightforwardly compared to resultsf rom the computational approaches. Gas-phase experimentsw ould offer considerable benefits in this context, since results could be directly linked to computational results withoutd eploying the very advanced calculations that are necessary to properly account for solventeffects on the photochemistry. [22] In this paper,w ed emonstrate an ew method that has the potentialt oc ontributes ignificantly to progress in this area by applyingU V-VIS laser photodissociation spectroscopy to a knownp hotoCORM, TryptoCORM( Scheme 1), to map its photochemistry at ap reviously unprecedented level of detail. [23][24][25] This is the first time that the photochemical properties of a photoCORM have been investigated in the gas phase, thus providing crucial experimental data to evaluate recentt heoretical strategies against.
Our experimental technique involves transferring the photo-CORM from solution into the gas-phase via electrospray ionization and then using novel Laser-Interfaced Mass Spectrometry (LIMS) to measuret he gaseous absorption spectrum along with all of the ionic photoproducts produced at each scanned wavelength. [26,27] Thus, we obtain ac omplete pictureo ft he excited states along with the accompanying photodissociation pathways. We complement these measurements with Cryogenic Ion Vibrational Spectroscopy (CIVS), [28] an advanced experimental technique that provides structural characterization of gaseous ions.
By directly measuring the wavelength dependence of CO ejectionf rom TryptoCORM, we are able to demonstrate an ew approacht oa ssessing the CO releasing property of the photo-CORM.T his represents as ignificanta lternative to the current principal methodf or determining this property,n amely the myoglobin (Mb)a ssay. [29] Motterlini and co-workersw ere the first to report this assay,w hich has subsequently been refined by other research groups. [30,31] The technique measuresC Or elease into solution indirectly through its conversion of deoxy-Mb to Mb-CO. The approach can be problematic,h owever, if CO loss from the CORM is reversible.I nc ontrast, the new LIMS approachp resented in this study provides ad irectp icture of the (non-reversible) amount of CO photoejected as af unction of wavelength, giving ac ompelling complementary approach to test Mb assays against andp roviding key information for selecting excitation wavelength versus amount of CO released.
TryptoCORM was the first visible-light activated CORM to exhibit ap otent effect against Escherichia coli. [24] One of its key advantages as ap otential therapeutic agent is that it shows low toxicityt owards mammalian cells and releases bio-benign Tryptophan on photoexcitationi na queous solution. [24] Trypto-CORM is thought to release up to 3C Om olecules upon aqueous-phase excitation at 400 nm (3.1 eV) dependent on conditions. [24] However, at the lower excitatione nergies (465 nm, 2.66 eV) that are preferred for medical applications, [32] fewer CO molecules are released (1.4 equivalent). The CH 3 CN group of TryptoCORM is labile, [33,34] so that the TryptoCORM-CH 3 CN complex( i.e. the TryptoCORM molecule minus the acetonitrile ligand)i sl ikely to exist in ad ynamic equilibrium with Trypto-CORM in solution,l eadingt oq uestionsa roundt he relative photochemical and hence medicinal effectiveness of the two species. In this work, we are able to directly compare the photochemistry of the TryptoCORM and TryptoCORM-CH 3 CN moieties forthe first time since our mass spectrometry based technique allows us to independently isolate both molecular species prior to photoexcitation.

Results and Discussion
Characterization of the gas-phase structures of electrosprayed TryptoCORM The protonated pseudom olecular ion, [MnL(CO) 3 (CH 3 CN)]·H + , where L = deprotonated Tryptophan, appeared strongly upon electrospray of TryptoCORM (Section S1). The most intense ion observed was the protonated molecular speciesw ithout the CH 3 CN ligand, that is, [MnL(CO) 3 ]·H + .W en ote that protonation of TryptoCORM is enhanced through the electrosprayp rocess, rather than [MnL(CO) 3 (CH 3 CN)]·H + being dominant in solution. [25] TryptoCORM and TryptoCORM-CH 3 CN have multiple possible protonation sites, with the mostl ikely ones indicated on Scheme 1. Protonation on the Mn metal center is also possible for TryptoCORM-CH 3 CN. [35] The protonation isomer(s) formed upon electrospray can be predicted by calculating the relative energies of the variousp rotonation site isomers or protomers. [36,37] For [MnL(CO) 3 (CH 3 CN)]·H + ,t he lowest-energy gasphase protonation site corresponds to the O1 position (Section S2), giving rise to the structuresd isplayed in Figures 1a and b. This pairofd iastereoisomers are predicted to be presentint he ratio of 86:14 (Section S2). Protonation at other sites leads to much higher-energy isomers that we predict are not populated.
Similarly for [MnL(CO) 3 ]·H + ,t he lowest-energy gas-phase structure ( Figure 1c:afolded structure) corresponds to O1 protonation, with other protomers lying at significantly higher energies (Table S5). The O1 protomer involvest he Tryptophan ligand folding to coordinate to the vacant metal center through the indole C=Cb ond (hapticity 2), at an interaction length (2.5 )t hat is similar to the Na tom of the CH 3 CN ligand in [MnL(CO) 3 (CH 3 CN)]·H + (2 ). (Table S4)  To directly test the extent of Tryptophan-Mn coordination, cryogenic ion vibrational spectroscopy was performed on the gaseous electrosprayed ions. [28,38] Figure 2d isplays the acquired IR spectrum of [MnL(CO) 3 ]·H + ,a long with the calculated spectra of the folded (Figure 2a Table 1l ists the calculated and experimental IR frequencies. Comparison of the experimental and computational spectra across the NH/OH stretchr egion (2400-3600 cm À1 )c onfirms that the computed spectrumf or the folded structure    3 ]·H + compared to (c) the experimentalI R spectrum overt he ranges 1600-2300a nd 2400-3600cm À1 .The calculated spectraa re scaled with respect to wavenumber by 0.94 over the 1600-2300 cm À1 region,a nd by 0.97 over the 2400-3600 cm À1 region.The calculated spectral intensities for the two ranges havebeen scaled so that the intensities of the most intensep eaks in each regiona re equal. Table 1. Calculated (CAM-B3LYP/Def2SV) and experimental vibrational frequencies for the folded and openc onformationali somerso f[ MnL(CO)3]·H + displayed in Figure 1.

Band
Folded n (cm À1 )Open n (cm À1 )Experimental n (cm À1 ) CO stretch region [ 3 ]·H + was also recorded across the CO stretching region (1600-2300 cm À1 ), with the experimental spectrum across this regiond isplaying one CO band in the carboxylic stretchr egion at 1694 cm À1 ,a longw ith three additional bands in the metal bound CO stretching region. The spectrumi sc onsistent with previous gaseous IR spectra of metal carbonyl complexes. [39][40][41][42] We note that bands II and III are only partially resolved. The experimental spectrum is again consistentwith the predicted spectrum for the folded structure ( Figure 2a), with the experimental bands occurring closer to the predicted vibrational frequencies for the folded structure than the open structure. IR-IR conformer-specific spectroscopy was performed to confirmt hat only as ingle isomer is present (SectionS4). [43] Thermal fragmentationpathways of [MnL(CO) 3

(CH 3 CN)]·H + + and [MnL(CO) 3 ]·H + +
Prior to performing laser photodissociation, we investigated the thermalf ragmentationp athways of [MnL(CO) 3 (CH 3 CN)]·H + and [MnL(CO) 3 ]·H + by performing higher-energy collisional dissociation( HCD) for energies between 0-30 %H CD. This experiment maps out the fragmentation pathwaysa saf unction of internal energy, [44] andi st herefore also of interesti nt he context of the known propensity of TryptoCORM to release CO upon thermale xposure, that is, to behavea sathermal-CORM. [9,23] The HCD data is included in Section S5, along with ad etailed discussion of the various HCD channels. Figure 3i llustrates the observed[ MnL(CO) 3 (CH 3 CN)]·H + fragmentation path-ways. The key observations to emerge from the HCD experimentsa re firstly,t hat the CH 3 CN ligand is very easily lost via collisional excitation,that is, the ion is metastable. [45] Secondly,n of ragmentation channel associated with loss of as ingle CO molecule is observed, and fragmentation with loss of two CO units is am uch less intense channel than loss of three COs. It therefore seems that the barrier to loss of the second CO is low once the first CO is lost, and similarly,t he barriert o3 CO loss is low once 2CO are ejected. [46,47] This situation may well be different in solution where solvation could modify these barrier heights.F inally,H CD can induce intramolecular fragmentation of Tryptophan at high collision energy. The HCD data for [MnL(CO) 3 ]·H + revealed that its fragmentation behavior closely mirrorst hat of [MnL(CO) 3 (CH 3 CN)]·H + , both in terms of the observedf ragmentation channels and their energy dependence.

(CH 3 CN)]·H + + and [MnL(CO) 3 ]·H + +
To gain insighti nto photo-induced loss of CO, we inspected the cationic photofragmentsa saf unctiono fl aser excitation wavelength. [MnL(CO) 3 (CH 3 CN)]·H + and [MnL(CO) 3 ]·H + were observed to photofragmenti nto an umber of distinctive channels, andthese are discussed in more detail in the next section. Since we are primarily interested in photo-triggered CO loss from TryptoCORM, we focus here on the major (most intense) photoinduced loss of CO channels, that is, pathways Comparison of the Figure5photofragment production spectra with the gaseous absorption spectra (Figure 4) clearly showst hat the propensity to eject CO closely mirrors the absorptions pectra. There are several additional points of note. CO ejectionf rom [MnL(CO) 3 ]·H + is significant over ag reater wavelength range than for [MnL(CO) 3 (CH 3 CN)]·H + .W hile this is most evident over the region from 380-300 nm, it is also significant in the key regionb etween 500-450nm. Importantly, the crosss ection for loss of 3CO falls by around 50 %o ng oing from 400-440 nm for [MnL(CO) 3 (CH 3 CN)]·H + butt he reduction is less for [MnL(CO) 3 ]·H + due to the increased electronic ab-  sorptioni nt he region. The primary CO production spectra therefore clearly show that incorporation of the labile CH 3 CN ligand into TryptoCORM optimizesl ight-triggered CO release furtheri nto the redder wavelength region. Ta ken together,t he CO production spectra (Figures 5a and 5b) also show that there is little enhancement in CO production on movingf rom 400 nm to bluer wavelengths such as 330 nm, and indeed total CO production dips on going through the 300 nm region.  (Figure 6c)r esults in strong production of [MnL(CH 3 CN)]·H + particularly over the low-energy region.T his product ion is striking as it is unanticipated on purely energetic grounds, given that HCD demonstrated that CH 3 CN is more readily lost compared to CO. As in the HCD experiment, no photofragment corresponding to ejection of as ingle CO is observed,a nd while loss of 2COs is observed (Figure 6d)i ti s negligible at higher energies.

Furtherdiscussion of TryptoCORM'sphotofragmentation profiles
An overview perspective on photofragment productioni s gained if we consider the quantum ion yields for the dominant photofragmentationc hannels of [MnL(CO) 3 (CH 3 CN)]·H + versus [MnL(CH 3 CN)]·H + .F igure 8d isplays these quantum ion yields as af unction of the entire wavelength range scanned. These plots strikingly show how the photofragment production is controlled by the morphology of the excited states, and not simply by the internal energy of molecules following photoexcitation, that is, photofragment production does not simply increasea safunctiono fp hotoexcitation energy.F or [MnL(CO) 3 (CH 3 CN)]·H + ,t he dominant channel which corresponds to loss of 3COs follows ar easonablys mooth bell curve shape that peaks at % 3.6 eV (344 nm). Across the 2.9-4.8 eV (320-258nm) region, the 3CO loss channel for [MnL(CO) 3 ]·H + displays av ery similarp rofile, again peaking at % 3.6 eV (344 nm). This clearly demonstrates that both species access the same excited-state surfacea cross the 2.9-4.8 eV (320-258 nm) region. At highere nergies, [MnL(CO) 3 (CH 3 CN)]·H + also increasingly loses its acetonitrile, while [MnL(CO) 3 ]·H + displays an increased propensity to lose the C 9 H 8 Nu nit. Both of these channels reflect the increasing internal energy of the complexes as af unction of photoexcitation.
The tryptophan C a -C b bond rupture fragment is characteristic of the dominant photofragmentt hat would be expected from gaseous protonated Tryptophan. [55] While thesee xcitations should dominate in the region above 4.3 eV (290 nm), the fact that the -C 9 H 8 Nl oss photofragmenti so bserved in the lower-energy bands is evidencet hat strong coupling occurs throughout the excitation range studied here. Indeed, this observation is in line with recent theoretical work from Fumanal et al. which has shown that the central metal atom plays a controlling role in the early time photophysics of transition metal complexes. [56] In their studies of am anganese (I) complex, ultrafastd ecay of the absorbing MLCT S 2 state was found to be mediated by vibronic coupling between the S 2 /S 1 and the upper singlet metal-centered and MLCT states. Crucially,i n the absence of strong spin-orbit coupling, S 2 !S 1 internal conversion was found to be indirecta nd mediated by distinctive upper electronic states.T his early time photophysics prepares the complexf or subsequent carbonyl dissociation.
The quantum ion yield plots provide clear evidence for the presence of three distinctive excitation regions for [MnL(CO) 3    the computed excited state surfaces map onto these quantum yield plots.

Implicationsf or PhotoCORM activityo fT ryptoCORM
Next, we turn to discussing our measurements of the direct photodecayp athways of isolated TryptoCORM in the context of its solution-phase behavior.I nitial myoglobin-based spectroscopic assays of TryptoCORM indicatedt hat the molecule did not release significant amountso fC Oi nt he dark, but released 2m olar equivalents of CO upon irradiation at 400 nm (3.1 eV). [24] However, subsequentw ork with leghemoglobin revealed that TryptoCORM did in fact release CO in the dark. This seeming contradictionw as explained as resulting from differences in the binding constanto fC Ot ot he molecule it was released from versus its binding constants to myoglobin and leghemoglobin. [24] This example highlights one of the known problemso fu sing myoglobin assays to measure CO release. [31] The technique demonstrated here provides an ew approach to determining the wavelength dependence of CO release from a potentialp hotoCORM in as traightforwarda nd unambiguous manner.W hile there will be known differences between the gaseous and solution-phase CO dissociation profile, [16] our results show that for TryptoCORM, lower molar equivalents of CO are produced (> 410 nm in the gas phase) in the visible region compared to the UV (< 400 nm in the gas phase). This result is entirely in line with solution-phase irradiation measurements. [24] It is important to acknowledge that theseo bservations do not imply that we believe that the gaseous dissociation mechanism mirrorst hat in solution.T he mechanisms of metal carbonyl photodissociation in solution are knownt o differ from the gas-phase. [57,58] For example, am etal carbonyl will typically eject ah ighern umber of carbonyl ligandsf ollowing UV/VIS excitation in the gas-phase compared to solution, since vibrational relaxation of the excesse nergy to the solvent bath is not availablef or gaseous systems. [58] Transienti nfrared spectroscopy of organometallic speciesi nt he gas phase has also providedanumber of propensity rules for linking those older gas-phase measurements with more recent solutionphase measurements. For example, it was observed that in the gas phase, that both the coordinatively unsaturated photofragment and the ejected ligand tend to be produced with more internal energy as the energy of the photolysis photon increases, and also that the nature of the electronic state accessed can influence branching ratios for products. [58] Ultra-fast time-resolved infra-red spectroscopy has very recently been used to demonstrate that, in solution,C O-dissociation proceeds to give 3 [MnL(CO) 2 (CH 3 CN)] as the dominantprocess on irradiation at 400 nm. [25] This occurs in under 1psa nd is followed by ac hangei ns pin and coordination of the solvent, S, to give 1 [MnL(CO) 2 (CH 3 CN)(S)] (t % 20 ps). Compared to the solution-phase photolysis measurements whichw ere conducted with two fixed wavelength diode LEDs, however,o ur photolysis experiment incorporates ab road scanning laser source.W et herefore obtain af ull picture of the spectralr egions where CO loss is maximized via loss of three CO units from the TryptoCORM, as showni ne xquisite detail on the quantum ion-yield plots of Figure 8. Perhaps just as importantly,t hese plots allow us to trace the extent to which maximum CO loss is compromised at the redder wavelengths, and hence make an informed decision as to the optimum wavelength for clinicalapplication.

Conclusions
In conclusion, we have measured the intrinsic absorption spectrum and photofragmentationp athways for TryptoCORM as an isolated molecular complex in the gas-phase. The photofragment productions pectra reveal the optimum excitation wavelengthsf or maximizing CO ejection from ap hotoCORM for the first time, and therefore demonstrate as traightforwarda nd widelya pplicable new methodology for assessing photoinduced CO release. Importantly,o ur gas-phaseresultsc an be directly linked to the solution-phase properties of the systemb y comparing the band positions of the gaseous and solutionphase absorption spectra.T he experiments reported here result from applying novel laser-interfaced mass spectrometry within an adapted commercialm ass spectrometer,atechnique that has considerable broader potentialf or applied photochemicala nd photophysical studies. [27,59] In particular,t he experiments performed here are not limitedt op hotoCORMs but could be readily applied to other photo-activated prodrugs including platinum anticancer complexesa nd H 2 So rC S 2 releasing therapeutical agents. [60][61][62] Finally,w en ote that our results also representt he first direct gas-phase measurement of the wavelength-dependentp hotochemical branching ratios (quantum ion yields) for at ransition metal carbonyl complex. The spectra presented here therefore provide an ew benchmark against whichh igh-level exciteds tate calculations of this challenging group of molecular systems can be compared.

Experimental Section
Cryogenic ion vibrational spectroscopy:T he gas-phase infrared photodissociation spectra presented here were obtained in a custom-built spectrometer described in detail elsewhere. [38] Briefly, solutions of TryptoCORM in CH 3 CN ( % 10 À3 m)w ith trace amounts of formic acid were electrosprayed to generate [MnL(CO) 3 ]·H + ions, where the Ll igand is deprotonated Tryptophan. Ions were directed to a3 Dq uadrupole ion trap held at 10 Kb yaclosed-cycle helium cryocooler,a nd D 2 -tagged adducts, that is, [MnL(CO) 3 ]·H + ·D 2 ,w ere then produced by introducing a % 1msb urst of He, seeded with 10 %D 2 .T he [MnL(CO) 3 ]·H + ·D 2 aggregates were then ejected into the time-of-flight mass spectrometer,m ass selected via ag ated deflector,and intersected with the output of aNd:YAGpumped tunable OPO/OPAl aser between the ranges 1400-2300 and 2400-3800 cm À1 .R esonant absorption of one IR photon leads to loss of the D 2 tag, that is, [Eq. (1)] ½MnLðCOÞ 3 ÁH þ Á D 2 þhn IR !½MnLðCOÞ 3 ÁH þ þD 2 ð1Þ thermal fragmentation characteristics of [MnL(CO) 3 (CH 3 CN)]·H + and [MnL(CO) 3 ]·H + ,u sing an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA, U.S.A.) as described previously. [44] HCD energies from 0-30 %H CD were employed. Solutions of TryptoCORM (1 10 À5 mol dm À3 )i nH 2 O:CH 3 CN 1:1w ere introduced to the mass spectrometer (for HCD and LIMS) through electrospray ionization using ac apillary temperature of 140 8C. Laser-interfaced mass spectrometry:G as-phase UV photodissociation experiments were conducted in al aser-interfaced amaZon ion-trap mass spectrometer (LIMS), which was modified as described in detail elsewhere. [26,63] The LIMS instrument has all the advantages of acommercial mass spectrometer (e.g. flexible ion sources, mass selection and isolation of primary and secondary ions and fragments via MS n schemes), coupled with the ability to record UV absorption and photofragmentation spectra in ar outine manner.T he UV photons in these experiments were produced by an Nd:YAG (10 Hz, Surelite) pumped OPO (Horizon) laser,g iving % 1mJa cross the range 2.1-5.3 eV (234-580 nm). Al aser step size of 1nmw as employed for all scans, with photofragmentation experiments being run with an ion accumulation time of 100 ms. A fragmentation time of 100 ms was employed, so that each massselected ion packet interacted with one laser pulse to minimize multiphoton excitation. Laser power studies were conducted at 400 and 280 nm to verify that only as ingle photon is required to induce photofragmentation.
Photodepletion intensity (PD) and photofragment production (PF) were calculated using Equations (2) and (3): Photofragmentation Intensity ¼ ð where Int ON and Int OFF are the peak intensities with laser on and off, Int Frag the fragment intensity with laser on, l the excitation wavelength (nm) and Pt he laser pulse energy (mJ). The photodepletion spectrum is considered to be equivalent to the gaseous absorption spectrum in the limit where fluorescence is negligible. Quantum ion yields are calculated according to Equation (4): where Int PFT is the sum of the photofragment ion intensities obtained with the laser on.
TryptoCORM synthesis: d-TryptoCORM was synthesized following apreviously published protocol. [24] Quantum chemistry calculations:D ensity Functional Theory (DFT) calculations were performed in Gaussian 09, [64] using standard methods. Full details are given in Sections S2 and S3, with optimized geometries in Section S4.